import numpy as np
import tensorflow as tf
from tensorflow import keras


def norm_pix(image, label):
    return image / 255, label


directoryTRAIN = "D:/Image_dataset/Image_datasetJPGnewBnW/Image_datasetJPGnewBnW/train"
directoryVALIDATION = "D:/Image_dataset/Image_datasetJPGnewBnW/Image_datasetJPGnewBnW/validation"

train_ds = tf.keras.utils.image_dataset_from_directory(directoryTRAIN,
                                                       seed=123, batch_size=16,
                                                       image_size=(200, 200), color_mode='rgb')#color_mode='grayscale')


val_ds = tf.keras.utils.image_dataset_from_directory(directoryVALIDATION,
                                                     seed=123, batch_size=16,
                                                     image_size=(200, 200), color_mode='rgb')#color_mode='grayscale')


train_ds = train_ds.map(norm_pix)
val_ds = val_ds.map(norm_pix)

model = keras.Sequential([
    #keras.layers.Conv2D(64, (3, 3), activation='relu', input_shape=(200, 200, 1)),
    keras.layers.Conv2D(64, (3, 3), activation='relu', input_shape=(200, 200, 3)),
    keras.layers.MaxPool2D((2, 2)),

    keras.layers.Conv2D(128, (3, 3), activation='relu'),
    keras.layers.MaxPool2D((2, 2)),

    keras.layers.Conv2D(256, (3, 3), activation='relu'),
    keras.layers.MaxPool2D((2, 2)),

    keras.layers.Flatten(),
    keras.layers.Dense(1024, activation='relu'),
    keras.layers.Dense(4, activation='softmax')
])

print(model.summary())

model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])

trainHistory = model.fit(train_ds, epochs=8, validation_data=val_ds)

#model.save("model.h5")
model.save("model_color.h5")

(loss, accuracy) = model.evaluate(val_ds)
print(loss)
print(accuracy)